JPH077959A - Controlling method for pwm inverter - Google Patents

Abstract

PURPOSE:To improve output characteristics of a PWM inverter by predicting a capacitor current of one sample ahead from a detected amount and an output voltage pattern of a present time point, and making the current follow up a command value. CONSTITUTION:A system having a PWM inverter and an AC LC filter so outputs a voltage pattern of the inverter that currents of filter capacitors 11-13 follow up a command value, and controls voltages of the capacitors 11-13. The currents of the capacitors 11-13 are predicted according to information such as an output amount detected at each sampling time, a present voltage pattern, etc. This predicting control is conducted to eliminate the influence of a calculating time of a digital control system. In the case of a periodic nonlinear load, sampling data of one period of a load current is recorded, and repeatedly controlled. Thus, output characteristics of the inverter are improved.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION A voltage type inverter to which the control method of the present invention is applied is not affected by the operation time of the control system,
Since the filter capacitor current can follow the command value, high-performance control characteristics can be obtained. Therefore, the present invention can be effectively applied to the fields of industrial machines, home electric appliances, etc. using the voltage source inverter.

[0002]

2. Description of the Related Art Various control methods have been devised for voltage type PWM inverters depending on the application. FIG. 2 is an example of a PWM voltage type inverter control method used conventionally, and FIG. 3 is its waveform diagram.

As shown in FIG. 3, voltage command values v _u ^* , v
_v ^*, _v is compared with the _w ^* and the triangular wave v _c, make a signal of the base drive circuit through the hysteresis comparator.
The output voltage amplitude value can be changed by adjusting the amplitude value of the voltage command value. At the top of FIG. 3, a triangular wave v _c having a height V _c, the phase voltage command values v _u ^*, v
_v ^*, _v shows a waveform diagram of the status of the comparison with the _w ^*, v _uN, v
_vN, v _wN each represents a voltage waveform between the neutral point of each _{phase, v uv, v vw, v} wu shows the voltage waveforms of respective phases, E _d is the voltage of the DC power source. Each of the inverters has a switching element 2 to which diodes are connected in antiparallel.
7 (eg transistors) in a three-phase bridge connection, two in series with virtual neutral points E _d / 2 respectively
DC voltage E _d is supplied to the DC input terminals from the DC power sources 1 and 1'having a voltage of 1. Connect 13 in parallel to supply AC output to the load.

[0004]

In the three-phase voltage source inverter shown in FIG. 2, the output voltage is determined by the command value and the triangular wave, but a considerable load fluctuation occurs with respect to the DC input voltage fluctuation and the load fluctuation. Although the error can be reduced by introducing feedback control, it may cause new problems such as stability. Further, the ripple of the DC input voltage directly affects the output voltage. Further, as can be seen from FIG. 2, since the three phases are independently controlled, it is difficult to reduce the ripple of the output voltage without increasing the switching frequency.

On the other hand, there has been devised a method of directly calculating the output voltage pattern of the inverter as a voltage vector by using the instantaneous space vector. By performing so-called instantaneous control, the control characteristics of the inverter are improved. However, in the case of performing instantaneous control, since complicated calculation is required, a microprocessor or digital signal processor (D
SP) is often used. In that case, the detection value is always delayed by the sampling time due to the influence of the calculation time, and the control performance is deteriorated.

[0006]

In a system including a PWM inverter and an AC LC filter, a voltage pattern of the inverter is output so that the filter capacitor current follows the command value, and the filter capacitor voltage is controlled. The filter capacitor current is predicted based on the output amount detected at each sampling time and information such as the current voltage pattern. By performing so-called predictive control, the influence of the operation time of the digital control system is eliminated. The periodic non-linear load can be solved by recording the sampling data for one cycle of the load current and repeating the control.

The details of the principle of the output voltage control method based on the instantaneous space voltage vector were first described by the applicant in 1993.
It is disclosed in Japanese Patent Application No. 5-26646 “PWM Inverter Control Method” filed on the 16th of March. Here, the concept is applied below.

FIG. 1 is a principle diagram of the present invention, in which (a) is P
The structure of a WM inverter system is shown, (b) has shown the control principle. The same reference numerals as those in FIG. 2 indicate the same parts, and the control circuit controls the base drive circuit to drive the inverter. In FIG. 1 (b) which is a principle diagram, θ _S is an angle at which the current vector should rotate in one sampling time, and ω is a rotation angular frequency.

Control Principle of Output Voltage The output amount of the PWM inverter can be converted into three-phase / two-phase, and the current voltage can be treated as a base. Also, as is generally known, a three-phase voltage source inverter has eight voltage vectors, including two zero voltage vectors. The output voltage / current vector equation of FIG.

[Equation 1] Is obtained as. v is an arbitrary output voltage vector, v _c is a capacitor voltage vector, and i, i _c and i _L are a filter reactor current vector, a capacitor current vector and a load current vector, respectively.

Considering the three-phase sinusoidal phase output voltage on the two-phase coordinates, a circular locus with a constant speed is drawn. Therefore, the fundamental current of the load capacitor must also have a circular locus.

Now, the sampling period is defined by the following equation.

[Number 2] _{T S = T / k (2} ) where, T is the fundamental period of the inverter, k is the sampling number of one cycle. That is, the sampling time T
_The current vector within _S may move along the circle of the command value.

Next, the switching area corresponding to the six non-zero voltage vectors is divided into six spatial area sectors every 60 degrees as shown in FIG. The output voltage vector corresponding to each sector is defined in the table below.

[Table 1]

The current can be made to follow the command value within the sampling time by using the two non-zero voltage vectors and the two zero voltage vectors corresponding to each sector.

From equation (1),

[Equation 3] However, t _o + t _i + t _j + t ₇ = T _s , i, j (i, j
= 1-6) is the number of the non-zero voltage vector. Δi _L
Is the amount of change in the load current and may be ignored if there is no sudden change in load.

For example, in the sector I, the voltage vector is V ₇ → V ₆ → V ₁ → V ₀ or V ₀ → V ₁ → V ₆ → V ₇ as shown in FIG. As described above, the output is performed in the determined period, and the capacitor current can follow the command value within one sampling time.

The selection of the zero voltage vector is performed by switching V ₁ , V ₃ , and V ₅ from the viewpoint of reducing the number of switching operations.
₀ is used, and V ₇ is used to switch between V ₂ , V ₄ , and V ₆ .

Capacitor Current Prediction Control Since an operation time is required to calculate the voltage vector to be output and its time width, the actual detection value is always delayed by the sampling time. In particular, since the filter capacitor current changes at high frequencies, a large error occurs due to the influence of the calculation time. Therefore, the current one sampling ahead is predicted from the information of the detected value and the voltage vector at the present time.

According to equation (1), the predicted current value is

[Equation 4] Is obtained as. The value with the subscript 0 added to the equation (4) represents the current value.

Output Period of Voltage Vector Next, a method of determining the time width of the output voltage vector of the inverter will be described.

The current command value is expanded on the α-β two-phase coordinate,
It is given by the following equation.

[Equation 5] The subscripts a and b in the above equation represent components on the α-β coordinate.
The current amplitude command value I _C ^* is obtained by using the voltage amplitude command value V _C ^* .

[Equation 6]

When the equation (4) is expanded on the α-β coordinate, the following equation is obtained.

[Equation 7] Where V is the amplitude value of the output voltage vector, C _a ,
C _b is a constant. By the above formula,

[Equation 8] Is obtained. t ₀ and t ₇ must be decided so as to minimize the current ripple, but here, for simplicity,

[Equation 9] Can be

In a transient state, t _i or t _j may become negative. In such a case, the reverse voltage vector may be output. Furthermore, in the case of a sudden load change, it may not be able to follow within one sampling time.

[Equation 10] It is also possible that In such a case, the voltage vector may be selected so that the current error (Δi _ca ² + Δi _cb ² ) ^1/2 is minimized.

The current error vector is

[Equation 11] Next to

[Equation 12] When the equations (11) and (12) are solved so that the error current vector is minimized, their t _{i and} t _j are equation (13).

[Equation 13] Therefore, the output voltage vector may be output at each predetermined time.

According to the equation (4), since the predictive calculation formula of the filter capacitor current also includes the change amount of the load current, even when the load current changes suddenly, the influence of the calculation time causes The calculation accuracy becomes poor. Therefore, in a periodic non-linear load such as a rectifier load, the load current can be recorded for one cycle and these data can be used in the next cycle to accurately predict the capacitor current.

[0025]

In the digital control system of the PWM inverter, the output current of the PWM inverter can be improved by predicting the capacitor current one sampling ahead from the current detected amount and the output voltage pattern and making the capacitor current follow the command value. .

[0026]

[Embodiment] The principle described above is applied to CVCF (constant voltage constant frequency).
An example applied to an inverter will be specifically shown. The circuit configuration is the same as that shown in FIG. The load is a Y-connection three-phase resistance load or a three-phase rectifier load.

The digital control circuit, as shown in FIG.
It is composed of an SP (digital signal processor) and an A / D converter. Here, three voltage inputs (v _C ), two current inputs (i _C ), and three current inputs (i _L ) A
The / D converter is shown. The A / D converter is controlled by the DSP, and in addition to the DC input voltage V _dc , the three-phase load current i _L , the capacitor current i _C , and the capacitor voltage v _C are input to the DSP as digital quantities every sampling time. The DSP executes a control algorithm based on the above-mentioned control principle at high speed.

In the embodiment, the fundamental frequency of the output voltage is 60 Hz
Then, the sampling frequency is set to 10.8kHz, and in that case, the switch frequency in the steady state is 5.4kHz. Good results were obtained for 100% sudden load change and rectifier load.

[0029]

The present invention predicts the filter capacitor current one sampling ahead using the current sampling detection value and the voltage output vector, and makes the prediction value follow the command value, so that the output voltage is calculated in the calculation time. Can be controlled without being affected by. Therefore, the waveform of the output voltage is significantly improved, and the control accuracy can be significantly improved.

[Brief description of drawings]

FIG. 1 is a principle diagram of the present invention, in which (a) is a diagram showing a configuration of a PWM inverter and (b) is a diagram showing a control principle thereof.

FIG. 2 is a principle diagram of an example of a control method of a PWM inverter that has been conventionally used.

FIG. 3 shows a waveform chart of each part of FIG.

FIG. 4 is a diagram showing spatial domain sectors corresponding to voltage vectors.

5 is a diagram showing an output sequence of voltage vectors in a region I. FIG.

[Explanation of symbols]

1 DC power supply 2 to 7 Switching element 8 to 10 Filter reactor 11 to 13 Filter capacitor v _uN , v _vN , v _wN Voltage waveform between each phase and neutral point v _uv , v _vw , v _wu Voltage between each phase Waveforms v _u ^* , v _v ^* , v _w ^* Voltage command value v _c Triangular wave waveform V _c Triangular wave height _Ed is DC voltage V ₀ , V ₇ zero voltage vector V _{1 to} V ₆ non-zero voltage vector v Arbitrary output voltage vector v _c Capacitor voltage vector i Filter reactor current vector i _c Capacitor current vector i _L Load current vector T _S Sampling time T Inverter fundamental wave period k Number of samplings in one period I _C ^* Command value of current amplitude V _C ^* Voltage amplitude command value C _a , C _b constant V _dc DC input voltage Δi _L Change in load current

Claims (1)

[Claims] 1. When digital control is performed in a system composed of a three-phase voltage source inverter and an AC LC filter, output voltage and current are detected at every sampling time, and a current detection value and a PWM output pattern are detected. The filter capacitor current one sampling ahead is predicted from the information, and the optimum two non-zero voltage vectors and two zero voltage vectors of the eight output voltage vectors of the inverter are used, and the filter capacitor is determined within the sampling time. A method for controlling a PWM inverter, characterized in that the predicted value current of (1) is made to follow the command value current and the output voltage (voltage of the filter capacitor) is controlled without the influence of the calculation time.